The eight agents behind QuantShift AI.

Each agent specialises in one job. An orchestrator routes your question to the right subset — only running what's needed. Every agent is industry-aware, evidence-grounded, and produces structured output that feeds into the final PDF report.

Layer 1 — Sensing

Continuously gather signals from the market, your customers, and the broader competitive landscape.

Market Signal Agent

Scans industry news, competitor moves, technology shifts, regulatory changes, and consumer sentiment.

When it runs: Always runs first when you ask 'what's changing in my market?'

Customer Job Agent

Applies Christensen's Jobs-to-be-Done framework to your customer data — uploaded PDFs, CSVs, or Notion pages.

When it runs: Runs when you ask 'why are customers actually buying from us?'

Disruption Pattern Agent

Watches for the five classic disruption signatures: low-end, new-market, performance-oversupply, business-model innovation, platform shifts.

When it runs: Runs alongside Market Signal during weekly briefings and full snapshots.

Layer 2 — Analysis

Make sense of the signals and turn them into structured strategic understanding.

Modularity & Interdependence Agent

Maps the value chain to identify where to invest in-house (interdependent, premium-margin) vs outsource (modular, commoditised).

When it runs: Runs when you ask about operations, supply chain, or build-vs-buy.

Competitive Position Agent

Models your position vs competitors with structured outputs: trajectory, margin pressure, differentiation strength, openings.

When it runs: Runs for any decision question and during full snapshots.

Hypothesis Agent

Generates 3–5 testable strategic hypotheses, each grounded in upstream agent outputs, with evidence for and against.

When it runs: Runs after analysis layers complete; feeds into Test Designer.

Layer 3 — Action

Decide and adapt — turn understanding into concrete moves.

Test Designer Agent

Converts each hypothesis into a low-cost, time-boxed market test (landing page, pricing test, interviews, pilot, ad test, survey).

When it runs: Runs after the Hypothesis Agent. Always pairs one test per hypothesis.

Adaptation Agent

When unexpected changes happen, recommends hold / pivot / accelerate / defer / investigate-further with specific action items.

When it runs: Runs when you ask 'X just happened, what do I do?' or during continuous monitoring.